I find the answer!

Since each thread maintains its own connection, your database must support at
least as many simultaneous connections as you have worker threads.


The development server creates a new thread for each request it handles,
negating the effect of persistent connections.


ref link:
https://github.com/django/django/blob/2ee21d9f0d9eaed0494f3b9cd4b5bc9beffffae5/docs/ref/databases.txt


【 在 wuhaochi (oo) 的大作中提到: 】
: 标  题: 如何处理django的数据库连接池？
: 发信站: 水木社区 (Thu Apr 10 17:32:42 2014), 站内
:
: django对每次request都发起一个connect。传说中django 1.6之前每次连接完了就关闭了， 到了django 1.6，终于有个CONN_MAX_AGE来指定连接存活时间。
: 但这有屁有啊!它还是每次request都发起连接请求。
: 在数据库里光看到连接数增长了，却不见django复用之前的连接……
:
: 按理不就如此，是不？那难道是我跟踪错了？
:
:     def connect(self):
:         """Connects to the database. Assumes that the connection is closed."""
:         # In case the previous connection was closed while in an atomic block
:         self.in_atomic_block = False
:         self.savepoint_ids = []
:         # Reset parameters defining when to close the connection
:         max_age = self.settings_dict['CONN_MAX_AGE']
:
:         print "wuhaochi", "connect to database", __file__
:
:         self.close_at = None if max_age is None else time.time() + max_age
:         self.errors_occurred = False
:         # Establish the connection
:         conn_params = self.get_connection_params()
:         self.connection = self.get_new_connection(conn_params)
:         self.init_connection_state()
:         if self.settings_dict['AUTOCOMMIT']:
:             self.set_autocommit(True)
:         connection_created.send(sender=self.__class__, connection=self)
:
:
:
:     def get_new_connection(self, conn_params):
:         return Database.connect(**conn_params)
:
: 最后的Database，即我绑定的postgresql adapter，到这一步，是肯定连过去了。
:
: 而self.connection 则在 读取 self.get_new_connection(conn_param)前根本没有判断。
:
:
: --
:
: ※ 来源:·水木社区 newsmth.net·[FROM: 116.247.85.*]
--
FROM 116.247.85.*